InsurePro: Predicting Insurance Charges¶

By: Arsh Vohra¶

pexels-pixabay-47327%20%281%29.jpg

Description:¶

Harness the power of machine learning to predict insurance charges with our advanced ML model. By leveraging a variety of data, including features such as age, BMI, number of children, and region, our model can accurately estimate insurance charges for individuals.

To ensure optimal performance, we employ a data preprocessing step that involves scaling the numeric columns. Scaling is essential to bring the numeric features to a similar scale and prevent any single feature from dominating the learning process. By applying scaling techniques such as standardization or normalization, we ensure that the features have comparable ranges and enable our ML model to make accurate predictions.

In [1]:
import pandas as pd
import numpy as np
import matplotlib
import matplotlib.pyplot as plt
import seaborn as sns
sns.set_style('darkgrid')
matplotlib.rcParams['font.size']=14
matplotlib.rcParams['figure.facecolor']='#00000000'
import opendatasets as od
import plotly.express as px
from sklearn.linear_model import LinearRegression
from sklearn.model_selection import train_test_split
from sklearn.metrics import mean_squared_error
from sklearn.preprocessing import StandardScaler
In [2]:
link='https://raw.githubusercontent.com/stedy/Machine-Learning-with-R-datasets/ff933238d9c30da179b7e3ad4b6ca938d67efc53/insurance.csv'
In [3]:
od.download(link)

Using downloaded and verified file: .\insurance.csv
In [4]:
df=pd.read_csv('insurance.csv')
In [5]:
df.head()
Out[5]:
age sex bmi children smoker region charges
0 19 female 27.900 0 yes southwest 16884.92400
1 18 male 33.770 1 no southeast 1725.55230
2 28 male 33.000 3 no southeast 4449.46200
3 33 male 22.705 0 no northwest 21984.47061
4 32 male 28.880 0 no northwest 3866.85520
In [6]:
(df.isna().sum()/df.shape[0])*100
Out[6]:
age         0.0
sex         0.0
bmi         0.0
children    0.0
smoker      0.0
region      0.0
charges     0.0
dtype: float64
In [7]:
df.info()

<class 'pandas.core.frame.DataFrame'>
RangeIndex: 1338 entries, 0 to 1337
Data columns (total 7 columns):
 #   Column    Non-Null Count  Dtype  
---  ------    --------------  -----  
 0   age       1338 non-null   int64  
 1   sex       1338 non-null   object 
 2   bmi       1338 non-null   float64
 3   children  1338 non-null   int64  
 4   smoker    1338 non-null   object 
 5   region    1338 non-null   object 
 6   charges   1338 non-null   float64
dtypes: float64(2), int64(2), object(3)
memory usage: 73.3+ KB
In [8]:
df.describe()
Out[8]:
age bmi children charges
count 1338.000000 1338.000000 1338.000000 1338.000000
mean 39.207025 30.663397 1.094918 13270.422265
std 14.049960 6.098187 1.205493 12110.011237
min 18.000000 15.960000 0.000000 1121.873900
25% 27.000000 26.296250 0.000000 4740.287150
50% 39.000000 30.400000 1.000000 9382.033000
75% 51.000000 34.693750 2.000000 16639.912515
max 64.000000 53.130000 5.000000 63770.428010

Univariate analysis¶

In [9]:
#Age
fig = px.histogram(df,x='age',marginal='box',nbins=64,title="Distribution Of Age",color='smoker')
fig.update_layout(bargap=0.1)
In [10]:
fig,axes = plt.subplots(1,2,figsize=(16,5))
fig.suptitle("Distribution of age")

sns.histplot(data=df,x='age',kde=True,ax=axes[0],hue='smoker')
sns.boxplot(data=df,x='age',ax=axes[1],y='smoker')

q25,q75=np.percentile(df['age'],25),np.percentile(df['age'],75)

IQR =q75-q25

upper = q75 + (1.5*IQR)
lower = q25 - (1.5*IQR)

out=0
for i in df['age']:
  if i > upper or lower > i:
    out=+1

print("{} Total Number of Outliers in Age:{}".format('\033[1m',out))

 Total Number of Outliers in Age:0
In [11]:
#BMI
fig = px.histogram(df,x='bmi',
                   marginal='box',
                   color_discrete_sequence=['red'],
                   title='Distribution of BMI'
                   )
fig.update_layout(bargap=0.1)
In [12]:
fig,axes=plt.subplots(1,2,figsize=(16,5))
fig.suptitle("Distribution Of BMI")

sns.histplot(data=df,x='bmi',ax=axes[0],color='r',kde=True)
sns.boxplot(data=df,x='bmi',ax=axes[1],color='r')


q25,q75 = np.percentile(df['bmi'],25),np.percentile(df['bmi'],75)

IQR = q75-q25

upper = q75 + (1.5*IQR)
lower = q25 - (1.5*IQR)

out = 0
for i in df['bmi']:
  if i > upper or lower > i:
    out+=1
print("{} Total Outliers in BMI:{}".format('\033[1m',out))

 Total Outliers in BMI:9
In [13]:
#Children
fig=px.histogram(df,x='children',
             marginal= 'box',
             color_discrete_sequence=['green'],
             title="Distribution of Children")
fig.update_layout(bargap=0.1)
In [14]:
fig,axes=plt.subplots(1,2,figsize=(16,5))

sns.histplot(data=df,x='children',kde=True,color='g',ax=axes[0])
sns.boxplot(data=df,x='children',color='g',ax=axes[1])

fig.suptitle("Distribution of Children")

q25,q75 = np.percentile(df.children,25),np.percentile(df.children,75)

IQR = q75-q25

upper = q75 + (1.5*IQR)
lower = q25 - (1.5*IQR)

out=0

for i in df['children']:
  if i > upper or i < lower:
    out+=1

print("{} Total number of outerliers in Children:{}".format('\033[1m',out))

 Total number of outerliers in Children:0
In [15]:
#Charges
fig = px.histogram(df,x='charges',
                   color='smoker',
                   marginal='box',
                   color_discrete_sequence=['red','green'],
                   title = ("Distribution of Charges"))
fig.update_layout(bargap=0.1)
In [16]:
fig,axes=plt.subplots(1,2,figsize=(16,5))

sns.histplot(data=df,x='charges',hue='smoker',kde=True,ax=axes[0])
sns.boxplot(data=df,x='charges',y='smoker',ax=axes[1])

fig.suptitle("Distribution Of Charges")

q25,q75 = np.percentile(df.charges,25),np.percentile(df.charges,75)

IQR = q75-q25
upper = q75 + (1.5*IQR)
lower = q25 - (1.5*IQR)

out=0

for i in df['charges']:
  if i > upper or i < lower:
    out+=1

print("{} Total number of outliers in Charges:{}".format('\033[1m',out))

 Total number of outliers in Charges:139
In [17]:
px.histogram(df,x='sex',title="Distribution of Sex")
In [18]:
fig,axes=plt.subplots(1,2,figsize=(16,5))
fig.suptitle("Distribution of Sex")

sns.countplot(df,x="sex",ax=axes[0])
df['sex'].value_counts().plot(kind='pie',autopct='%0.2f%%',shadow=True,startangle=50,explode=[0.01,0.01])
Out[18]:
<Axes: ylabel='sex'>
In [19]:
px.histogram(df, x='smoker', color='sex', title='Distribution of Smoker with respect to Sex')
In [20]:
plt.figure(figsize=(16,5))
plt.title("Distribution of Smoker with respect to Sex")

sns.countplot(df,x='smoker',hue='sex')
Out[20]:
<Axes: title={'center': 'Distribution of Smoker with respect to Sex'}, xlabel='smoker', ylabel='count'>
In [21]:
plt.figure(figsize=(16,5))
plt.title("Distribution of Charges with respect to Region")
a= df.groupby(['region'])[['charges']].sum()

sns.barplot(a,y='charges',x=a.index,)
Out[21]:
<Axes: title={'center': 'Distribution of Charges with respect to Region'}, xlabel='region', ylabel='charges'>
In [22]:
px.scatter(df,x='age',y='charges',color='smoker',title="Charges Vs Age")
In [23]:
plt.figure(figsize=(16,5))
plt.title("Charges Vs Age")
sns.scatterplot(df,x='age',y='charges',hue='smoker')
Out[23]:
<Axes: title={'center': 'Charges Vs Age'}, xlabel='age', ylabel='charges'>
In [24]:
px.scatter(df,x='bmi',y='charges',color='smoker',title="Charges Vs BMI")
In [25]:
plt.figure(figsize=(16,5))
plt.title("Charges Vs BMI")

sns.scatterplot(df,x='bmi',y='charges',hue='smoker')
Out[25]:
<Axes: title={'center': 'Charges Vs BMI'}, xlabel='bmi', ylabel='charges'>
In [26]:
plt.figure(figsize=(10,5))
sns.heatmap(df.corr(),annot=True,fmt='0.2f',cmap='magma',linewidths=1,center=1)

C:\Users\Arsh\AppData\Local\Temp\ipykernel_26888\1168954917.py:2: FutureWarning:

The default value of numeric_only in DataFrame.corr is deprecated. In a future version, it will default to False. Select only valid columns or specify the value of numeric_only to silence this warning.
Out[26]:
<Axes: >
In [27]:
df['smoker'].replace({'no':0,'yes':1},inplace=True)
df['sex'].replace({'female':0,'male':1},inplace=True)
df= pd.get_dummies(df,columns=['region'])
In [28]:
x = df.drop(['charges'],axis=1)
In [29]:
y=df['charges']
In [30]:
xtrain,xtest,ytrain,ytest = train_test_split(x,y,train_size=0.7,test_size=0.3,random_state=42)
In [31]:
model = LinearRegression()
In [32]:
model.fit(xtrain,ytrain)
Out[32]:
LinearRegression()
In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook.
On GitHub, the HTML representation is unable to render, please try loading this page with nbviewer.org.
LinearRegression()
In [33]:
def rmse(y_true,y_pred):
    rmse = mean_squared_error(y_true,y_pred,squared=False)
    return rmse
In [34]:
print("Accuracy on Training data is :{}".format(model.score(xtrain,ytrain)))
print("Accuracy on Testing data is :{}".format(model.score(xtest,ytest)))
print("RMSE on Train data is :{}".format(rmse(ytrain,model.predict(xtrain))))
print("RMSE on Test data is :{}".format(rmse(ytest,model.predict(xtest))))

Accuracy on Training data is :0.742398009060026
Accuracy on Testing data is :0.7696118054369009
RMSE on Train data is :6142.519818737067
RMSE on Test data is :5812.100272258874
In [35]:
for i,j in enumerate(xtrain.columns):
  print("Weight of {} is {}".format(j,model.coef_[i]))

Weight of age is 261.29692414191965
Weight of sex is 104.81182298826343
Weight of bmi is 348.9069151601751
Weight of children is 424.11912829161696
Weight of smoker is 23628.367222359924
Weight of region_northeast is 596.056589237782
Weight of region_northwest is 109.12197876483239
Weight of region_southeast is -374.91224932883875
Weight of region_southwest is -330.2663186737813
In [36]:
w=pd.DataFrame(model.coef_)
w.rename({0:"Weights"},axis=1,inplace=True)
w.rename({0:xtrain.columns[0],1:xtrain.columns[1],2:xtrain.columns[2],3:xtrain.columns[3],4:xtrain.columns[4],5:xtrain.columns[5],6:xtrain.columns[6],7:xtrain.columns[7],8:xtrain.columns[8]},axis=0,inplace=True)
w.reset_index(inplace=True)
w.rename(columns={w.columns[0]: 'Feature'},inplace=True)
w
Out[36]:
Feature Weights
0 age 261.296924
1 sex 104.811823
2 bmi 348.906915
3 children 424.119128
4 smoker 23628.367222
5 region_northeast 596.056589
6 region_northwest 109.121979
7 region_southeast -374.912249
8 region_southwest -330.266319
In [37]:
numeric_cols =['age','bmi','children']
scaler = StandardScaler()
scaler.fit(df[numeric_cols])
Out[37]:
StandardScaler()
In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook.
On GitHub, the HTML representation is unable to render, please try loading this page with nbviewer.org.
StandardScaler()
In [38]:
scaler.mean_
Out[38]:
array([39.20702541, 30.66339686,  1.09491779])
In [39]:
scaler.var_
Out[39]:
array([197.25385199,  37.16008997,   1.45212664])
In [40]:
scaled_inputs = scaler.transform(df[numeric_cols])
In [41]:
numeric_df = pd.DataFrame(scaled_inputs).rename({0:"age",1:"bmi",2:"children"},axis=1)
numeric_df
Out[41]:
age bmi children
0 -1.438764 -0.453320 -0.908614
1 -1.509965 0.509621 -0.078767
2 -0.797954 0.383307 1.580926
3 -0.441948 -1.305531 -0.908614
4 -0.513149 -0.292556 -0.908614
... ... ... ...
1333 0.768473 0.050297 1.580926
1334 -1.509965 0.206139 -0.908614
1335 -1.509965 1.014878 -0.908614
1336 -1.296362 -0.797813 -0.908614
1337 1.551686 -0.261388 -0.908614

1338 rows × 3 columns

In [42]:
categorical_cols = ['sex','smoker','region_northeast','region_northwest','region_southeast','region_southwest']
categorical_df = df[categorical_cols]
In [43]:
categorical_df
Out[43]:
sex smoker region_northeast region_northwest region_southeast region_southwest
0 0 1 0 0 0 1
1 1 0 0 0 1 0
2 1 0 0 0 1 0
3 1 0 0 1 0 0
4 1 0 0 1 0 0
... ... ... ... ... ... ...
1333 1 0 0 1 0 0
1334 0 0 1 0 0 0
1335 0 0 0 0 1 0
1336 0 0 0 0 0 1
1337 0 1 0 1 0 0

1338 rows × 6 columns

In [44]:
new_df = pd.concat([numeric_df,categorical_df],axis=1)
In [45]:
inputs = new_df
target = df['charges']
In [46]:
x_train,x_test,y_train,y_test = train_test_split(inputs,target,train_size=0.7,test_size=0.3,random_state=42)
In [47]:
scaled_model = LinearRegression()
scaled_model.fit(x_train,y_train)
Out[47]:
LinearRegression()
In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook.
On GitHub, the HTML representation is unable to render, please try loading this page with nbviewer.org.
LinearRegression()
In [48]:
print("Accuracy on Training data is :{}".format(model.score(xtrain,ytrain)))
print("Accuracy on Testing data is :{}".format(model.score(xtest,ytest)))
print("RMSE on Train data is :{}".format(rmse(ytrain,model.predict(xtrain))))
print("RMSE on Test data is :{}".format(rmse(ytest,model.predict(xtest))))

Accuracy on Training data is :0.742398009060026
Accuracy on Testing data is :0.7696118054369009
RMSE on Train data is :6142.519818737067
RMSE on Test data is :5812.100272258874
In [49]:
for i,j in enumerate(inputs.columns):
  print("Weight for {} is:{}".format(j,scaled_model.coef_[i]))

Weight for age is:3669.8392723470292
Weight for bmi is:2126.9043303283493
Weight for children is:511.08143575240115
Weight for sex is:104.81182298826657
Weight for smoker is:23628.367222359902
Weight for region_northeast is:596.0565892377891
Weight for region_northwest is:109.12197876484359
Weight for region_southeast is:-374.91224932884296
Weight for region_southwest is:-330.26631867377785
In [50]:
pd.DataFrame({
    "Feature":np.append(numeric_cols + categorical_cols,1),
    "Weight": np.append(scaled_model.coef_,scaled_model.intercept_)
})
Out[50]:
Feature Weight
0 age 3669.839272
1 bmi 2126.904330
2 children 511.081436
3 sex 104.811823
4 smoker 23628.367222
5 region_northeast 596.056589
6 region_northwest 109.121979
7 region_southeast -374.912249
8 region_southwest -330.266319
9 1 8437.701437
In [51]:
new_w=pd.DataFrame(scaled_model.coef_)
new_w.rename({0:"Weights"},axis=1,inplace=True)
new_w.rename({0:inputs.columns[0],1:inputs.columns[1],2:inputs.columns[2],3:inputs.columns[3],4:inputs.columns[4],5:inputs.columns[5],6:inputs.columns[6],7:inputs.columns[7],8:inputs.columns[8]},axis=0,inplace=True)
new_w.reset_index(inplace=True)
new_w.rename(columns={new_w.columns[0]: 'Feature'},inplace=True)
new_w
Out[51]:
Feature Weights
0 age 3669.839272
1 bmi 2126.904330
2 children 511.081436
3 sex 104.811823
4 smoker 23628.367222
5 region_northeast 596.056589
6 region_northwest 109.121979
7 region_southeast -374.912249
8 region_southwest -330.266319
In [52]:
plt.rcParams['font.size']=30
fig,axis = plt.subplots(1,2,figsize=(55,15))
sns.barplot(w,x='Weights',y='Feature',ax=axis[0])
axis[0].set_title("Model without Scaling")
sns.barplot(new_w,x='Weights',y='Feature',ax=axis[1])
axis[1].set_title("Model after Scaling")
fig.suptitle("Model Comparison")
fig.set_dpi(300)
In [53]:
px.bar(w,x='Weights',y='Feature',title='Model without Scaling')
In [54]:
px.bar(new_w,x='Weights',y='Feature',title='Model after Scaling')